We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.SD

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Sound

Title: Tonal harmony, the topology of dynamical score networks and the Chinese postman problem

Abstract: We introduce the concept of dynamical score networks for the representation and analysis of tonal compositions: a score can always be interpreted as a dynamical network where every chord is a node and each progression links successive chords. We demonstrate that in a tonal harmony context this network displays scale-free properties, and optimal (most economical) chord progressions can be found by solving a path optimization like the Chinese postman problem. Moreover, the dynamical network can be viewed as a time series of a non-stationary signal and as such can be partitioned for the automatic identification of key regions using well-established techniques for time series analysis and change point detection. Based on this interpretation we introduce a key-finding algorithm that does not rely on comparisons with pre-determined reference sets, as in the Krumhansl-Schmuckler model, or extensive corpora, as in machine-learning approaches. Finally, we show how the principles discussed in this work can be used to design a generative model of tonal compositional design.
Comments: 21 pages
Subjects: Sound (cs.SD); Social and Information Networks (cs.SI); Audio and Speech Processing (eess.AS)
Cite as: arXiv:2006.01033 [cs.SD]
  (or arXiv:2006.01033v1 [cs.SD] for this version)

Submission history

From: Marco Buongiorno Nardelli [view email]
[v1] Mon, 1 Jun 2020 15:59:56 GMT (3928kb)

Link back to: arXiv, form interface, contact.